Research On High-speed Milling Tool Wear Status Identification Based On Wavelet Neural Network

High speed milling (HSM) technology was more and more applied in Manufacturing such as aerospace,mould,automobile,etc.The using of HSM can significantly improve the productivity, improve quality and reduce the cost.However,HSM tool wear fast and service life is short."Cutter breakage"phenomenon easily appears if not for real-time monitoring. Therefore,the tool wear monitoring had important practical significance for high speed milling machining technology application.The dissertation is on the base of high speed milling of hardened steel experiment. First,the milling force and coefficient model were established,as well as a new coefficient method—principal component to regression analysis the model of coefficient was used. Through the simulation test's establish,simulation results show that the simulation signal established by the model was very close to the measured signal,and it can be used as a neural network training signal. Then,wavelet packet was used to decompose and de-noise the milling force signal,as well as extracts the signals'energy characteristics as import vectors of BP Neural Network.Based on the strong nonlinear mapping and classified capability of Neural Network,this dissertation adopted the combination of wavelet package analysis and BP(Back Propagation) Neural Network to identify the tool wear state. Through the establishment of BP network structure,training and test samples was structured for training and testing,the result showed that the network can accurately identify tool wear state. It laid a good foundation for the high speed milling cutter real-time monitoring.